trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9874476987447699,
  "eval_steps": 500,
  "global_step": 59,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.016736401673640166,
      "grad_norm": 7.772676690434981,
      "learning_rate": 8.333333333333333e-08,
      "logits/chosen": -2.807276487350464,
      "logits/rejected": -2.7759768962860107,
      "logps/chosen": -315.42626953125,
      "logps/rejected": -227.5915985107422,
      "loss": 0.6931,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "epoch": 0.16736401673640167,
      "grad_norm": 7.445230697975675,
      "learning_rate": 4.930057285201027e-07,
      "logits/chosen": -2.755582094192505,
      "logits/rejected": -2.746222972869873,
      "logps/chosen": -271.9510192871094,
      "logps/rejected": -260.6771240234375,
      "loss": 0.6915,
      "rewards/accuracies": 0.5069444179534912,
      "rewards/chosen": 0.005189303774386644,
      "rewards/margins": 0.003351197810843587,
      "rewards/rejected": 0.001838105614297092,
      "step": 10
    },
    {
      "epoch": 0.33472803347280333,
      "grad_norm": 6.71762655270701,
      "learning_rate": 4.187457503795526e-07,
      "logits/chosen": -2.786973476409912,
      "logits/rejected": -2.7680816650390625,
      "logps/chosen": -261.3984069824219,
      "logps/rejected": -250.85269165039062,
      "loss": 0.6745,
      "rewards/accuracies": 0.7093750238418579,
      "rewards/chosen": 0.03283267468214035,
      "rewards/margins": 0.04093491658568382,
      "rewards/rejected": -0.008102240040898323,
      "step": 20
    },
    {
      "epoch": 0.502092050209205,
      "grad_norm": 7.3796516085644255,
      "learning_rate": 2.8691164100062034e-07,
      "logits/chosen": -2.787944793701172,
      "logits/rejected": -2.76890230178833,
      "logps/chosen": -293.7830810546875,
      "logps/rejected": -254.1843719482422,
      "loss": 0.6487,
      "rewards/accuracies": 0.6781250238418579,
      "rewards/chosen": -0.02327023446559906,
      "rewards/margins": 0.12699303030967712,
      "rewards/rejected": -0.1502632349729538,
      "step": 30
    },
    {
      "epoch": 0.6694560669456067,
      "grad_norm": 8.12956318519775,
      "learning_rate": 1.4248369943086995e-07,
      "logits/chosen": -2.760505199432373,
      "logits/rejected": -2.737980604171753,
      "logps/chosen": -264.1119384765625,
      "logps/rejected": -255.556884765625,
      "loss": 0.6353,
      "rewards/accuracies": 0.684374988079071,
      "rewards/chosen": -0.11192457377910614,
      "rewards/margins": 0.16053779423236847,
      "rewards/rejected": -0.2724623680114746,
      "step": 40
    },
    {
      "epoch": 0.8368200836820083,
      "grad_norm": 8.537872275866219,
      "learning_rate": 3.473909705816111e-08,
      "logits/chosen": -2.7704710960388184,
      "logits/rejected": -2.750438690185547,
      "logps/chosen": -278.9338073730469,
      "logps/rejected": -287.13104248046875,
      "loss": 0.6201,
      "rewards/accuracies": 0.621874988079071,
      "rewards/chosen": -0.21925131976604462,
      "rewards/margins": 0.1369343101978302,
      "rewards/rejected": -0.3561856746673584,
      "step": 50
    },
    {
      "epoch": 0.9874476987447699,
      "step": 59,
      "total_flos": 0.0,
      "train_loss": 0.6490467928223691,
      "train_runtime": 1626.347,
      "train_samples_per_second": 9.397,
      "train_steps_per_second": 0.036
    }
  ],
  "logging_steps": 10,
  "max_steps": 59,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}